An MPEG(moving picture experts group)-2 video includes three picture types: an intra-coded (I) picture, a bidirectionally-coded (B) picture and a predictive-coded (P) picture. The MPEG-2 video is coded in frame picture units or field picture units. The I picture can be decoded regardless of other pictures, the P picture can be decoded from a previous I or P (I/P) picture, and the B picture can be decoded from previous I/P pictures and subsequent I/P pictures.
When the I, B and P pictures are all included in input pictures, the sequence of decoding differs from that of displaying. Accordingly, timing between decoding and displaying of the input pictures must be properly controlled to restore the original picture. Therefore, the video decoder must include frame memories for decoding the I or P pictures. Further, a memory must be provided wherein the previous and subsequent I/P frames for B picture decoding are stored.
When the B picture is a frame picture that must be displayed in field units, the B picture should be displayed after at least one field is stored. However, when the B picture is stored by only one field, one frame must be decoded while displaying one field. As a result, decoding time distribution is not efficient.
In addition, where one frame of the B picture is decoded and stored, and the size of a memory for B picture storage is set as two frames to store one following frame of the B picture during displaying the stored B picture of one frame, the timing between decoding and displaying is easily controlled, but hardware burden is increased.
Since the maximum frame size of an MPEG-2 main profile high level is about 27 Mbit, it is economical to set the memory space for B picture storage as one frame. However, when the B picture is stored only by one frame and the B picture is a frame picture which should be displayed in field units, the control of timing between decoding and displaying of successive B pictures becomes complicated.